Financial Models for Laboratory

Decision Making

Robert Schmidt, MD, PhD, MBA

University of Utah Department of Pathology
ARUP Laboratories

Suzanne Carasso, MBA, MT (ASCP)

ARUP Laboratories

March 19, 2015

 Introduction
Laboratory personnel are periodically confronted with complex
decisions such as buy versus lease, add a new test to the menu or
bring a reference test in-house. Such decisions are often made with
simple models that do not adequately capture risk, incorporate
alternative courses of action, or allow for sequential decisions that
evolve over time. As a result, decision makers often obtain
suboptimal results.

In this webinar, cutting edge techniques that incorporate risk,

facilitate the comparison of multiple alternatives, and provide insight
into common laboratory decisions will be presented. Attendees will
receive training in building financial models using Microsoft Excel and
Palisade Decision Tools, a popular add-in.

Participants will learn to use decision trees and simulation models

and then apply their knowledge to analyze whether to perform a test
in house or send it to a reference laboratory.

2
 Learning Objectives
• Determine when advanced modeling techniques
are likely to be helpful

• Explain how simulation models are used to

incorporate risk analysis in decisions

• Build simple models using Excel add-ins to

analyze problems using decision trees and
simulation

3
Session Faculty
Robert Schmidt, MD, PhD, MBA
• Medical Director, ARUP Laboratories

• Areas of Expertise
o Quantitative Analysis/Modeling
o Clinical Epidemiology
o Operations Management
o Diagnostic Testing
 Cost Effectiveness Analysis
 Meta-Analysis
 Literature Evaluation
 Laboratory Utilization

• Past Life
o Assistant Professor, Operations Management, University of Minnesota
o Associate Professor, Operations Management, University of Southern
California

4
Session Faculty
Suzanne Carasso, MBA, MT (ASCP)
Director, Business Solutions Consulting, ARUP Laboratories

• Consulting Director, ARUP Laboratories

• Areas of Expertise:
o Healthcare strategies for transitioning from volume to
value based care
o Laboratory legal structure and business models
o Value analysis and development of lab value proposition
o Strategy/business planning
o Market, operations and financial analyses

• Education
o B.S. Medical Technology, University of Tennessee
o MBA, University of Colorado at Denver

5
The purpose of this webinar is to educate participants
to make better decisions in the clinical and anatomic
pathology laboratory using financial models and risk-
based analysis.

• Understand financial models

• Analyze risk
• Demonstrate tools for risk analysis

6
What is a financial model?

7
Financial Model

8
Financial Model

9
 Financial Models

• Always wrong
• Sometimes useful

10
 Examples of “Wrong” Models

• Ideal gas laws

• Newtonian fluids
• Laws of motion (ignore friction, point masses)
• Perfect competition

11
 How are models useful?

• Eliminate bad ideas

• Provide insight
o Relationships between variables
o Uncertainty

• Provide predictions
o Don’t need to be perfect
o “fit for use”

12
 Simple Example

Cost = Labor + Reagents + Overhead

13
What about uncertainty?

14
 Distribution of Total Cost
Total / Cost
15,557 23,446

5.0% 90.0% 5.0%

2.0

1.8

1.6

1.4

Total / Cost
1.2
Minimum 13,899.79
Values x 10^-4

1.0 Maximum 26,328.91

Mean 19,100.15

0.8 Std Dev 2,412.86

Values 1000

0.6

0.4

0.2

0.0

15
Old Way (point estimate)

16
New Way (probabilistic estimate)

17
 How to do it (continued)

• Open Excel
• Click the @Risk Toolbar

18
How to do it (continued)

19
How to do it (continued)

20
 How to do it (continued)
Enter minimum, most likely, maximum

21
How to do it (continued)

22
How to do it (continued)

23
How to do it (continued)

24
How to do it (continued)
Set the iterations to 1000
Click “Start Simulation”

25
 Voila!
Total / Cost
15,557 23,446

5.0% 90.0% 5.0%

2.0

1.8

1.6

1.4

Total / Cost
1.2
Minimum 13,899.79
Values x 10^-4

1.0 Maximum 26,328.91

Mean 19,100.15

0.8 Std Dev 2,412.86

Values 1000

0.6

0.4

0.2

0.0

26
Each input has a distribution

27
Repeat calculations 1,000 times
obtain inputs from distributions

28
 The Question:

How can we apply this theory to a

realistic laboratory scenario?

29
 The Answer:

Create a realistic scenario.

30
The Scenario:
Build a financial model using a sales forecast and
five-year proforma to determine the rate at which the
laboratory sales team will capture the attainable
market

The Process:
• Define inputs for sales forecast
• Identify sources of uncertainty in sales forecast
• Develop 5-year forecast and financial projections
• Evaluate net present value
• Analyze one-way sensitivity analysis: Tornado Diagram
• Analyze two-way sensitivity analysis: Strategy Map

31
Sales Forecast requires five inputs
• Total Available Market (TAM)
• Attainable Market Share (AMS)
• Current Market Share (CMS)
• Sales Rate per person year (SRPY)
• Number of sales persons (N)

32
 Sales Forecast – Sources of Uncertainty

Sales Forecast Inputs Input Value Minimum Most Likely Maximum

Total Available Market (TAM) $8,000,000 $7,000,000 $8,000,000 $9,000,000
Attainable Market Share (AMS) 60% 50% 60% 70%
Current Market Share (CMS) 30% 25% 30% 35%
Sales Rate per person year (SRPY) $73,333 $50,000 $70,000 $100,000
Number of sales people (N) 6
YEAR 1: Sales per year (SRPY x N) $440,000

Salary and Operation Inputs Input Value Minimum Most Likely Maximum
Investment Cost $ 60,000 $ 40,000 $ 60,000 $ 80,000
Total Salary Expense/yr $ 366,000 $ 50,000 $ 60,000 $ 73,000
Total Operating Expense/yr $ 3,000,000 $ 1,000,000 $ 3,000,000 $ 5,000,000
Revenue Growth Rate: 5% 3% 5% 7%
Inflation Rate 3% 2% 3% 5%
Discount Rate 15%

33
 Pro-Forma Financial Statement
Estimated Financial Pro-forma
Revenue Year 1 Year 2 Year 3 Year 4 Year 5
Current Revenue $ 2,400,000 $ 2,400,000 $ 2,400,000 $ 2,400,000 $ 2,400,000
Incremental New Sales Revenue $ 440,000 $ 902,000 $ 1,387,100 $ 1,896,455 $ 2,400,000
Total Sales Revenue (Current + New Sales Revenue) $ 2,840,000 $ 3,302,000 $ 3,787,100 $ 4,296,455 $ 4,800,000 $19,025,555

Expenses Year 1 Year 2 Year 3 Year 4 Year 5

Initial Investment $ (60,000) - - - - - ($60,000)
Total Salary Expense $ 366,000 $378,200 $390,807 $403,834 $417,295 $1,956,135
Total Operating Expense $ 3,000,000 $3,100,000 $3,203,333 $3,310,111 $3,420,448 $16,033,893
Total Expenses $ (60,000) $3,366,000 $3,478,200 $3,594,140 $3,713,945 $3,837,743 $17,930,027

Cash Flow $ (60,000) ($526,000) ($176,200) $192,960 $582,510 $962,257 $975,528

Return Analysis
Discount Rate 15%
NPV $287,715

34
 Estimated Net Present Value
NPV / Input Value
0.00 3.90

47.8% 40.2% 12.0%

1.4

1.2

1.0

NPV / Input Value

0.8
Minimum -$7,593,761.56
Values x 10^-7

Maximum $8,389,886.03
Mean $172,304.32
0.6 Std Dev $3,014,563.66
Values 5000

0.4

0.2

0.0

Values in Millions ($)

2
0

8
4

10
-8

-6

-4

-2

35
 Tornado Diagram
(One-Way Sensitivity Analysis)
NPV / Input Value
Inputs Ranked By Effect on Output Mean

-$4,769,798.52 $5,195,022.45

-$699,482.41 $906,611.75

-$649,010.12 $892,027.53

-$512,617.62 $826,658.85

-$237,434.60 $491,461.23

3
-$152,671.25 $478,907.79

-$140,451.47 $459,300.16

-$95,631.96 $440,518.28

-$56,968.12 $320,178.87

$49,288.50 $397,619.25
Baseline = $172,304.32

1
NPV / Input Value

5
3
2
0

6
4
-5

-4

-3

-2

-1
Values in Millions ($)

36
 Strategy Map
(Two Way Sensitivity Analysis)

37
 Value of Information

Uncertainty in Sales Forecast

• Driven by Uncertainty in sales
• Market Research reduces uncertainty
• How to evaluate?

38
Decision Scenario

39
Value of Information

40
 Value of Information

Analyst Analyst 2 (well connected)

1

41
 Value of Perfect Information
What is the most you would pay Analyst 2?

Value of Perfect Info = 4,910,000 – 4805000 = 105,000

42
How to build a decision tree

43
Give the tree a name
Click OK

44
Name the Decision “Buy Info”
Name the branches yes and no

45
Right click on upper terminal node
Click “node settings”
Change to chance

46
 Right Click chance node
Add branch
Rename branches high medium low

47
Link to the probabilities and payoffs
probabilities above the line
payoffs below
Use absolute references (click F4)

Probability

Payoff

48
 Right click on the chance node
Copy Subtree
Right click on end node of “Yes” Branch
Paste subtree

49
 Change the payoffs on the lower subtree to zero
change each end node on the lower subtree to a “go vs no
go” decision
add payoffs

50
Voila!

51
Predicting the Impact of the FDA ruling on LDTs

52
Classification Probability

53
Cost of Approval Process

54
 Distribution of Approval Cost
Total
8.31 20.59

5.0% 90.0% 5.0%

1.4

1.2

1.0

Total
0.8
Minimum $1,944,314.67
Values x 10^-7

Maximum $27,797,931.07
Mean $14,264,953.03
0.6 Std Dev $3,749,277.22
Values 1000

0.4

0.2

0.0

Values in Millions ($)

5
0

10

15

20

25

30
55
Should we perform this test in-house?

56
 Difference (In-house vs send-out)
Difference / actual
-5,760 10,124

5.0% 90.0% 5.0%

1.0

0.9

0.8

0.7

Difference / actual
0.6
Minimum -11,030.78
Values x 10^-4

0.5 Maximum 16,410.98

Mean 1,188.67
0.4 Std Dev 4,918.53
Values 1000

0.3

0.2

0.1

0.0
0

57
 Quick Review

• Financial Modeling
o Risk Analysis
 Uncertainty in inputs
 Uncertainty in outputs

o Identify Risk Drivers

o Value of Information

58
 Is it worth the trouble?
• Easy to do
• Gain insight
o Focus on the important stuff
o Ignore the trivia

• Manage Risk
o Identify weak spots
o Develop options

• Increase Value

59
 Sources for Simulation Software

• Crystal Ball (Oracle)

• @Risk (Palisade)
• Risk Solver Pro (Frontline Systems)
• Many others

60
 Discussion:
Where can this be applied?

What problems would you like to see solved?

61
 Summary

• Financial Modeling Adds Value

o Can be applied to many problems
o Simple tools are available
• We would like to know:
o What risky decisions do you make?

62
© 2015 ARUP Laboratories